MIGRATION OF ADULT SOCKEYE SALMON PAST COLUMBIA RIVER DAMS, THROUGH RESERVOIRS AND DISTRIBUTION INTO TRIBUTARIES, 1997

Transcription

1 Technical Report 24-1 MIGRATION OF ADULT SOCKEYE SALMON PAST COLUMBIA RIVER DAMS, THROUGH RESERVOIRS AND DISTRIBUTION INTO TRIBUTARIES, 1997 A report for Project MPE-P-95-1 by G.P. Naughton, M.L. Keefer, T.C. Bjornn, M.A. Jepson, C.A. Peery, K.R. Tolotti, R.R. Ringe, and P.J. Keniry U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit University of Idaho, Moscow, ID and L.C. Stuehrenberg National Marine Fisheries Service 2725 Montlake Blvd, East, Seattle, Washington for U.S. Army Corps of Engineers Portland and Walla Walla Districts and Bonneville Power Administration Portland, Oregon 24

2 Preface Recent studies of adult salmon and steelhead migrations past dams, through reservoirs, and into tributaries with radio telemetry began in 199 with planning, purchase and installation of equipment for studies at the Snake River dams. Adult spring and summer chinook salmon were outfitted with transmitters at Ice Harbor Dam in 1991 and 1992, at John Day Dam in 1993 and reports of those studies are available (Bjornn et al. 1992; 1994; 1995; 1998; 1999; 23). The focus of adult salmon passage studies was shifted to the lower Columbia River dams in 1995 when telemetry equipment was set up at the dams and in tributaries. In this report we present information on the overall migration of sockeye salmon from release, past each of the dams in the Columbia River and into tributaries in Acknowledgments Many people assisted in the field work and data compilation for this project and the successful completion was made possible by Bob Dach and Teri Barila, the Corps of Engineers project officers at the time. Michelle Feeley, Brian Hastings, Megan Heinrich, Steve Lee, Mark Morasch and Jay Nance assisted in project operations and data processing and analysis. ii

3 Table of Contents Preface... ii Abstract... iv Introduction...1 General Methods...5 Monitoring Fish Movements...5 Outfitting Salmon with Transmitters...13 Receiver and Antenna Outages...15 Data Collection and Processing...2 Statistical Methods...23 Methods and Results...24 Passage, Migration History, and Final Distribution of Sockeye Salmon...27 Methods...27 Passage at Dams...28 Effects of Environmental Conditions on Sockeye Salmon Passage at Dams...31 Multivariate models...38 Effects of Injury on Passage Times...39 Passage Through Reservoirs...42 Passage Past Multiple Dams...44 Fallbacks at Dams...45 Effect of Injury on Fallback...53 Effect of Fallbacks on Passage Time...55 Reascension Over Dams, Escapement and Final Distribution After Fallbacks...58 Timing of Migration Past Dams and into Tributaries...61 Tag Dates for Specific Stocks of Sockeye Salmon with Transmitters...63 Reach Survival Estimates...63 Last Recorded Distribution of Sockeye Salmon with Transmitters...65 Recaptures of Sockeye Salmon with Transmitters...66 Fate of Sockeye Salmon with Transmitters...7 Discussion...74 References...78 iii

4 Abstract We captured 577 sockeye salmon Oncorhynchus nerka in the adult trapping facility at Bonneville Dam in 1997, released them with radio transmitters, and studied their passage past dams, through reservoirs and into tributaries. We set up radio receivers at Columbia and Snake river dams and at the mouths of major tributaries to monitor movements of salmon. Recaptures of salmon at hatcheries, weirs and traps, and data from mobile tracking were used to complete the migration history. We believe 57 fish retained transmitters beyond the release site and migrated upstream. Of the 57 fish, 1% returned to the Bonneville Dam tailrace and 98.6% were known to have passed the dam. Eighty-six percent of the 57 fish passed The Dalles Dam, 82% passed John Day Dam, 8% passed McNary Dam, 76% passed Priest Rapids Dam, 75% passed Wanapum Dam, 73% passed Rock Island Dam and 42% passed Rocky Reach Dam. Median times for sockeye salmon to pass individual Columbia River dams ranged from.3 d at The Dalles Dam to 1.4 d at Rocky Reach Dam. Median passage rates through reservoirs ranged from 36.4 km/d through the McNary pool to 64.7 km/d through the John Day pool. Median times to pass through reservoirs ranged from.6 d to 4.6 d. The median migration rate through the unimpounded Hanford Reach on the mid-columbia River was 28.2 km/d. From first passage of the tailrace at Bonneville Dam, median passage times past multiple dams were 6.9 d to the top of McNary Dam, 17.3 d to the top of Rock Island, and 19. d to the top of Rocky Reach Dam. In 1997, sockeye passed Bonneville Dam from late May through late August, with peak counts occurring in early and mid-july. Passage times for tagged fish at individual dams, were not strongly correlated with flow, spill, or turbidity. Cumulative passage times past multiple projects was negatively correlated with the date fish first passed the Bonneville Dam tailrace, with later migrating fish migrating at faster rates than those earlier in the migration. However, the relationship was weak with r 2 values <.3. Turbidity, spill, and flow at lower Columbia River dams explained relatively low proportions of the variability in passage times past multiple dams. The incidence of marine mammal injuries, descaling, and head injuries at time of tagging varied significantly during the migration. Injuries, however, appeared to have a limited impact on fish passage times. Marine mammal and descaling injuries also did not appear to affect fallback rates, but fish with head injuries fell back at dams at significantly higher rates than fish without head injuries. At least 164 sockeye salmon, 29% of the fish with transmitters that passed Bonneville Dam, fell back over or through Bonneville or other dams 181 times in Forty-three percent of all fallback events occurred at Bonneville Dam. One to seven percent of the fish that passed The Dalles, John Day and McNary dams fell back; 2 to 7% fell back at Priest Rapids, Wanapum, Rock Island and Rocky Reach dams. Fallbacks at any dam added to overall passage time past multiple dams. Using median passage times, one or more fallbacks at any dam added 1 to 7 days to overall passage time when compared to fish that did not fall back, differences, that were significant at lower Columbia River dams, but not at middle Columbia River Dams. Fish that fell back multiple times had the longest median passage times. iv

5 About 87% of sockeye salmon that fell back subsequently reascended all dams where they fell back. Of fish that did not reascend, about 27% subsequently entered tributaries downstream from the location of the fallback and probably did not reach spawning areas. From 63 to 1% of sockeye salmon that fell back at Columbia River dams eventually returned to tributary sites up- or downstream from the dam where they fell back. At most individual dams, sockeye salmon that fell back escaped to tributaries at significantly lower rates than fish that did not fall back. Migrations into individual tributaries were typically spread over 6 to 8 weeks. Because we did not monitor some mid-columbia River tributaries with fixed receivers, sockeye arrival at the first dam downstream was used as a surrogate for arrival at those sites. The median date sockeye salmon passed Rock Island Dam was 19 July for Wenatchee River stocks. The median first date at Wells Dam was 2 July for Methow and Okanogan river stocks, including those fish last recorded at Wells Dam. The median passage date at Bonneville Dam was 29 June for both Wenatchee and Okanogan river stocks. Reach survival estimates within the main stem Columbia/Snake river hydrosystem exceeded 96% for all sampled reaches. Reach survival estimates in the lower Columbia River were between 96% and 98% and estimates were > 97% through the mid-columbia River reaches. About 17% of tagged fish were reported recaptured in fisheries, at hatcheries, weirs or traps, at spawning grounds, or their transmitters were found along river corridors. Sixty-eight percent of reported recaptures were in tribal fisheries, 22% at spawning grounds, 7% at weirs or traps, and 3% in sport fisheries. About two-thirds of all recaptures were in the lower Columbia River and one-third was in the mid-columbia River basin. Our best estimate of the final fate for all radio-tagged sockeye salmon in 1997 was 2.8% downstream from Bonneville Dam, 18% between the top of Bonneville Dam and the McNary Dam tailrace, 5% between the top of McNary Dam to the Priest Rapids Dam tailrace, 37% in the Columbia River between the top of Priest Rapids Dam to Wells Dam, and 38% upstream from Wells Dam. Escapements were 68.5% in tributaries, 12.1% were reported recaptured in main stem tribal or sport fisheries and one fish (.2%) was reported captured in tributary sport fishery, 3.1% of transmitters were known or presumed regurgitated in non-spawning areas, and 16.1% were unaccounted for. Most notably, only a single sockeye salmon of 27 (3.7%) tagged at Bonneville Dam during the period 24 July 5 August successfully reached a spawning tributary. Fish that were unaccounted for may have been harvested but not reported to us, may have regurgitated transmitters that were not recovered or located, may have entered tributaries undetected, may have spawned at main stem locations, or may have died and were not detected as mortalities. The largest proportion of unaccounted-for fish (16.1%) were last recorded between the top of Rocky Reach Dam and the tailrace of Wells Dam. Another 15% were last recorded between the top McNary Dam and the tailraces of Priest Rapids and Ice Harbor dams. v

6 Introduction Studies of the passage of adult salmon Oncorhynchus spp. and steelhead O. mykiss at the lower Columbia River dams began in 1995 with the setup of radio telemetry equipment, and fish were outfitted with transmitters in 1996, 1997, 1998, 2, 21, and 22. In this report, we present information on passage of sockeye salmon at each of the dams, beginning with Bonneville Dam, and their migrations through reservoirs and into monitored tributaries throughout the basin in Sockeye salmon were only tagged during As in the previous studies, radio telemetry was used to monitor salmon movements at dams, up the rivers, and into tributaries. The study described herein was undertaken because of concerns of the U.S. Army Corps of Engineers (Corps), state and federal fish agencies and tribes, those expressed in section 63 of the Northwest Power Planning Council s (NPPC) 1987 Columbia River Basin Fish and Wildlife Program, and later reflected in the Biological Opinion on operation of the Federal Columbia River Power System, that studies were needed to ensure that passage of adult salmon and steelhead past the dams and through the reservoirs was as efficient as possible. Study plans were developed in consultation with Corps personnel, and with biologists in other federal, state, and tribal fish agencies. Public utility districts (PUDs) for Chelan and Douglas counties supplied one-third of the radio tags and maintained receiver sites at Rock Island, Rocky Reach and Wells dams. Research was conducted by personnel of the Idaho Cooperative Fish and Wildlife Research Unit (ICFWRU) and NOAA s National Marine Fisheries Service with logistical support, cooperation, and funding from the Corps, Bonneville Power Administration, US Geological Survey and PUDs. We set up receivers/antennas in 1997 at dams and tributaries in the lower Columbia River, at Priest Rapids and Wanapum dams on the mid-columbia River, at lower Snake River dams, at the lower end of the Clearwater River and Snake River near Asotin, WA, and at selected tributaries to the Clearwater and Salmon rivers (Figure 1). Receivers and antennas at mid-columbia River dams and tributaries upstream from Wanapum Dam were maintained by LGL Limited Environmental Associates for the Public Utility Districts of Douglas and Chelan counties (Alexander et al. 1998; English et al. 1998). LGL Limited also monitored radio-tagged fish in mid-columbia River tributaries. Fish with transmitters returned to tributaries, dams, traps, and hatcheries upriver from the uppermost fixed telemetry sites, and we used recaptures of those fish and data from mobile tracking to gain information about distribution of fish in tributaries. In 1997, sockeye salmon passed Bonneville Dam from late May through late August, with peak counts occurring in early and mid-july (Figure 2). Counts at lower Columbia River dams in 1997 were about 7 to 8% of the 1-year average (Table 1), about 9% of the 1-year average at Priest Rapids and Rock Island dams and greater than 1% at Rocky Reach and Wells dams. 1

7 Entiat Similkameen Canada U.S.A. Methow Distribution of radio receivers in 1997 Univ. of Idaho Receiver Site Wenatchee Chief Joseph Columbia River Wind WHR LWS Hood BO Deschutes Bonneville Dam - 34 receivers The Dalles Dam - 9 receivers John Day Dam - 9 receivers McNary Dam - 19 receivers Priest Rapids Dam - 1 receivers Klickitat TD Rock Island Priest Rapids Yakima John Day John Day Wells Rocky Reach Wanapum McNary Umatilla Grand Coulee LM LG Ice Harbor Walla W. Wanapum Dam - 12 receivers TUC Lower Granite Grande Ronde Imnaha Ice Harbor Dam - 11 receivers Lower Monumental Dam - 5 receivers Little Goose Dam - 5 receivers Lower Granite Dam - 5 receivers Little Salmon Snake River Clearwater River Lochsa SFS MFS Selway Salmon River Figure 1. Location of radio receivers at dams and major tributaries within the Columbia River study area in Does not include mid-columbia River and tributary sites maintained by Public Utility Districts for Chelan and Douglas counties. 2

8 3 2 Bonneville Dam n = 46,872 fish Avg. sockeye count sockeye count 1 3 The Dalles Dam n = 32,45 fish Number of sockeye salmon (thousands) John Day Dam n = 35,747 fish 3 2 McNary Dam n = 37,56 fish 1 1-Jun 1-Jul 1-Aug 1-Sep 1-Oct Figure 2. Number of adult sockeye salmon counted at Bonneville, The Dalles, John Day, McNary, Priest Rapids, Rock Island, Rocky Reach and Wells dams in 1997 with 1-year average counts (1987 to 1996). 3

9 3 2 Priest Rapids Dam n = 45,43 fish Avg. sockeye count sockeye count 1 3 Rock Island Dam n = 41,296 fish Number of sockeye salmon (thousands) Rocky Reach Dam n = 3,777 fish 3 2 Wells Dam n = 25,37 fish 1 1-Jun 1-Jul 1-Aug 1-Sep 1-Oct Figure 2. Cont. 4

10 Table 1. Adult sockeye salmon counted at main stem dams in 1997, and the 1997 counts as a percentage of the 1-year mean (1987 to 1996). Data from USACE Annual Fish Passage reports Percent of Dams Count 1-year Mean Bonneville 47,8 81 The Dalles 32,43 71 John Day 35,83 81 McNary 38,43 83 Priest Rapids 45,412 9 Rock Island 41,54 93 Rocky Reach 3, Wells 25, For much of the spring and summer of 1997, flow and spill in the Columbia and Snake rivers were nearly double the previous 1-year averages, and peak spill levels in May and June were often several times higher than average (1987 to 1996, Figure 3). Secchi disk visibility was well below average at all monitored dams in 1997 throughout the sockeye salmon migration (Figure 4). Water temperatures were slightly colder than the 1-year average at the lower Columbia and lower Snake river dams until late July, when they were near average; late summer water temperatures at Bonneville Dam were slightly warmer than average (Figure 5). Temperatures at Priest Rapids Dam were similar to average throughout the migration. This study in 1997 used radio telemetry on a large scale (577 sockeye salmon outfitted with radio transmitters) to assess the proportion of adult sockeye salmon that successfully passed dams in the lower and middle Columbia River, and their passage times at the dams and through reservoirs. Cumulative passage times and minimum escapements from Bonneville Dam past multiple dams were also estimated. The influence of environmental conditions on migration and fallback rates, relations between fallback and passage, final distributions for fallback and non-fallback salmon, and survival rates through reaches and to major tributaries were estimated for salmon tagged in General Methods Monitoring Fish Movements Radio telemetry was the primary means of assessing movements and passage rates of adult sockeye in the Columbia River in Priority dams for intensive study in 1997 were Bonneville, McNary, Ice Harbor, Priest Rapids, and Wanapum dams. They were fully outfitted with receivers and antennas to monitor all fishway entrances and exits, as well as the tailraces to determine when salmon with transmitters approached 5

11 Bonneville Dam Av Flow Av Spill 1997 Flow 1997 Spill The Dalles Dam 4 Mean daily flow and spill (kcfs) John Day Dam McNary Dam Apr 23-Apr1-May 15-May 1-Jun 6-Jun 28-Jun 1-Jul 2-Jul 1-Aug11-Aug 1-Sep Figure 3. Mean daily flow and spill volumes at Bonneville, The Dalles, John Day, McNary, and Priest Rapids dams in 1997 with 1-year averages (1987 to 1996). 6

12 Priest Rapids Dam Av Flow Av Spill 1997 Flow 1997 Spill 1 1-Apr 1-May 1-Jun 1-Jul 1-Aug 1-Sep Figure 3. Cont. dams. We also increased telemetry coverage from 1996 at The Dalles and John Day dams. In 1997, we set up receivers and antennas on all major tributaries upstream from Bonneville Dam (Figure 1 and Table 2). Receivers/antennas set up on tributaries were near the mouths, but far enough upstream so that transmitter signals from fish in the Columbia or Snake rivers would not be picked up and recorded. At some tributaries we installed receivers/antennas upstream or downstream from the tributary mouths to monitor salmon with transmitters in the main stem as they approached and proceeded upstream past a tributary. We also set up receivers and antennas to monitor passage at main stem sites at the Bridge of the Gods on the lower Columbia River and the Hanford Reach in the mid-columbia River. Additional receiver sites at dams and tributaries were maintained upstream from Wanapum Dam by Public Utility Districts for Douglas and Chelan counties (see Alexander 1998; English 1998). We monitored sockeye salmon movements with fixed-site radio receivers at dams and at the mouths of tributaries, and by mobile tracking in areas not covered by fixedsite antennas. Additional information was collected at upriver dams, traps and weirs and from fishers that returned transmitters. We used SRX receivers with Yagi antennas to determine when fish first entered the tailrace area of a dam. Digital spectrum processors (DSP) added to SRX receivers could simultaneously monitor several frequencies and antennas; DSPs were particularly helpful in monitoring movements of adults into and through fishways at dams. SRX/DSP receivers were connected to underwater antennas made of coaxial cable and were positioned near all fishway entrances, exits, and inside fishways at dams where fish were monitored intensively. We also used SRX receivers connected to Yagi 7

13 1 5 Bonneville Dam Gray: Av Secchi depth 1997 Secchi depth 1 The Dalles Dam 5 Secchi depth (ft) John Day Dam McNary Dam 5 1 Priest Rapids Dam 5 Av Secchi Depth Apr 15-Apr 29-Apr 13-May 27-May 1-Jun 24-Jun 8-Jul 22-Jul 5-Aug 19-Aug Figure 4. Mean daily Secchi disk visibility at Bonneville, The Dalles, John Day, McNary, and Priest Rapids dams in 1997 with 1-year averages (1987 to 1996). antennas near the mouths of main stem tributaries, at previously mentioned main stem sites, and on tributaries (Figure 1). For more details on receiver and antenna installation and the evolution of monitoring techniques for the adult passage project, see Bjornn et al. (1998; 2d). Three trucks were outfitted with 4-element Yagi antennas and SRX receivers to track fish in areas not covered by fixed-site receivers. Two boats were similarly outfitted to facilitate mobile tracking in reservoirs, as well as the free-flowing section of the 8

14 Water temperature (deg C) Bonneville Dam The Dalles Dam John Day Dam McNary Dam Priest Rapids Dam Gray: Av Water Temp 1997 Water Temp 1-Apr 15-Apr 29-Apr 13-May 27-May 1-Jun 24-Jun 8-Jul 22-Jul 5-Aug 19-Aug Figure 5. Mean daily water temperature at Bonneville, The Dalles, John Day, McNary, and Priest Rapids dams in 1997 with 1-year averages (1987 to 1996). Columbia River between Pasco and Priest Rapids Dam. In 1997, sections of the lower Columbia River were mobile-tracked approximately twice each month during the sockeye salmon migration. Segments of the Wind, White Salmon, Little White Salmon, Klickitat, and Deschutes rivers were also mobile-tracked occasionally. Additional tributaries, including those upstream from Priest Rapids Dam and downstream from Bonneville Dam were mobile tracked by cooperating agencies, primarily during the fall of

15 Table 2. Location of receivers at dams and tributaries in 1997, with site codes, number and type of aerial (A) and underwater (U) antennas at each site, description of site, and river kilometers from Columbia River mouth for some sites. Location Site Code Antennas Type Site description Bonneville Dam 1BO 1 A Tailrace, south side 2BO 1 A Tailrace, north side 3BO 1 A Downstream end of navigation lock 4BO 3 U Powerhouse 1, south end entrances 5BO 3 U Powerhouse 1, sluice gates 6BO 6 U Powerhouse 1, sluice gates 7BO 4 U Powerhouse 1, sluice gates 8BO 5 U Powerhouse 1, sluice gates ABO 1 U Top of Bradford Island ladder BBO 4 U South end of spillway ladder entrance CBO 4 U North end of spillway ladder entrance DBO 7 U Powerhouse 2, south shore entrances EBO 5 U Powerhouse 2, orifice gates FBO 4 U Powerhouse 2, orifice gates GBO 5 U Powerhouse 2, orifice gates HBO 5 U Powerhouse 2, orifice gates JBO 4 U Powerhouse 2, orifice gates KBO 5 U Powerhouse 2, orifice gates LBO 5 U Powerhouse 2, north shore entrances MBO 5 U North shore ladder transition pool NBO 4 U North shore ladder and transition pool OBO 3 U Washington ladder/umt channel junction PBO 1 U Top of Washington shore ladder QBO 3 U Top of navigation lock RBO 1 A Spillway forebay, facing north SBO 1 A Spillway forebay, facing south TBO 1 U Powerhouse 1, ice and trash sluiceway UBO 1 U Powerhouse 2, ice and trash sluiceway VBO 3 U A-Branch ladder transition pool WBO 3 U B-Branch ladder transition pool XBO 4 U Cascades Island ladder transition pool YBO 1 A Upstream from navigation lock ZBO 3 U UMT channel The Dalles Dam 1TD 1 A Tailrace, south side 2TD 1 A Tailrace, north side ATD 2 U South spillway entrance BTD 3 U West powerhouse entrance CTD 4 U South shore ladder entrance DTD 3 U South shore transition pool ETD 6 U North shore ladder entrance 5TD 1 U Top of Washington shore ladder FTD 1 U Top of Oregon shore ladder John Day Dam 1JD 1 A Tailrace, south side 2JD 1 A Tailrace, north side AJD 5 U Oregon shore ladder and transition pool BJD 3 U North powerhouse entrance CJD 6 U Washington shore ladder, transition pool DJD 2 U Wash. shore ladder, near diffuser pool 1

16 Table 2. Continued. Location Site code Antennas Type Site description EJD 2 U Oregon shore ladder, near diffuser pool 6JD 1 U Top of Oregon shore ladder 7JD 1 U Top of Washington shore ladder McNary Dam 1MN 1 A Tailrace, south side 2MN 1 A Tailrace, north side 3MN 3 U Oregon shore ladder entrance 4MN 6 U Oregon shore ladder transition pool 5MN 4 U Orifice gates 6MN 6 U Orifice gates 7MN 6 U Orifice gates 8MN 6 U Orifice gates 9MN 5 U Orifice gates AMN 5 U Orifice gates BMN 3 U North powerhouse entrance CMN 3 U Washington shore ladder entrance DMN 3 U Washington shore ladder transition pool EMN 1 U Top of Oregon shore ladder FMN 1 U Top of Washington shore ladder GMN 1 A Bottom of navigation lock HMN 5 U Top of navigation lock JMN 1 U Exit from juvenile bypass KMN 2 A Upstream end of juvenile bypass Priest Rapids Dam 1PR 1 A Tailrace, east side 2PR 1 A Tailrace, west side 4PR 5 U East shore ladder entrance 5PR 5 U Orifice gates 6PR 6 U Orifice gates 7PR 6 U Orifice gates 8PR 5 U West powerhouse entrance APR 1 U Top of West shore ladder BPR 6 U East shore ladder and transition pool CPR 1 U Top of East shore ladder Wanapum Dam 1WP 1 A Tailrace, east side 2WP 1 A Tailrace, west side 3WP 4 U East ladder entrance 4WP 4 U East ladder transition pool 5WP 3 U Orifice gates 6WP 4 U Orifice gates 7WP 5 U Orifice gates 8WP 3 U Orifice gates 9WP 3 U Orifice gates AWP 2 U West ladder entrance BWP 1 U Top of east shore ladder CWP 1 U Top of west shore ladder Ice Harbor Dam 1IH 1 A Tailrace, north side 3IH 4 U South shore ladder entrance 4IH 4 U Orifice gates 5IH 4 U Orifice gates 6IH 4 U Orifice gates 11

17 Table 2. Continued. Location Site code Antennas Type Site description 7IH 2 U North powerhouse entrance 8IH 4 U North shore entrance, transition pool, top 9IH 2 U Top of south shore ladder TIH 5 U South shore ladder transition pool 1CHAR 1 A Forebay, 3 km upstream from dam 2CHAR 1 A Forebay, 3 km upstream from dam Lower Monumental Dam 1LM 1 A Tailrace south side 2LM 4 U South shore ladder entrance, exit 3LM 4 U South powerhouse entrances 7LM 3 U North ladder entrance 8LM 1 U Top of north ladder Little Goose Dam 1GO 1 A Tailrace south side 2GO 4 U South shore ladder entrance 5GO 6 U North powerhouse entrances 6GO 4 U North shore entrance 7GO 1 U Top of south shore ladder Lower Granite Dam 1GR 1 A Tailrace, south side 6GR 6 U North powerhouse entrances 8GR 2 U Top of south shore ladder 1WI 1 A Forebay, 2 km upstream from dam 2WI 1 A Forebay, 2 km upstream from dam Bridge of Gods BOG 1 A RKM Wind River WIN 1 A River mouth (RKM 249.2) WNM 1 A River mouth (RKM 19.4) Little White Salmon R. LWS 1 A River mouth (RKM 261.) LWD 1 A Down Columbia from LWS (RKM 26.1) LWU 1 A Up Columbia of LWS (RKM 261.3) White Salmon River WHR 1 A River mouth (RKM 27.9) WHD 1 A Down Columbia from WHR (RKM 27.3) WHU 1 A Up Columbia from WHR (RKM 271.) Hood River HDR 1 A River mouth (RKM 272.6) Klickitat River KTR 1 A River mouth (RKM 29.7) Deschutes River DES 1 A River mouth (RKM 328.9) DSM 1 A Down Columbia from DES (RKM 327.1) SHF 1 A Sherars Falls (RKM 396.3) John Day River JDR 1 A River mouth (RKM 355.7) Umatilla River UMR 1 A River mouth (RKM 467.1) Walla Walla River WWR 1 A River mouth (RKM 56.) Yakima River YAK 1 A River mouth (RKM ~54) Hanford Reach HFL 1 A RKM HFR 1 A RKM Snake River SNR 1 A River mouth (RKM 762.3) Clearwater River CWR 1 A River mouth (RKM 753.3) SFC 1 A South Fork Clearwater River (RKM 867.6) Selway River SEL 1 A River mouth (RKM 96.1) Lochsa River LOC 1 A River mouth (RKM 93.8) Grande Ronde River GRR 1 A River mouth (RKM 794.7) Imnaha River IMR 1 A River mouth (RKM 867.6) Salmon River LSR 1 A Near Riggins (RKM 963.2) SFS 1 A South Fork (RKM 194.8) 12

18 Table 2. Continued. Location Site code Antennas Type Site description MFS 1 A Middle Fork (RKM ) USR 1 A Upper Salmon River (RKM 124.3) Sites maintained by Public Utility Districts of Douglas and Chelan counties Rock Island Dam RI Wenatchee River WEN River mouth TM1 Tumwater Dam Entiat River ENR River mouth Rocky Reach Dam RR Outfitting Salmon with Transmitters Radio transmitters were placed in 577 adult (no jacks) sockeye salmon trapped in the adult fish facility at Bonneville Dam in 1997 as they migrated upstream to natal streams or hatcheries. The salmon were transported to release sites at Dodson and Skamania Landings about 9.5 km downstream from Bonneville Dam. Tagging of adult sockeye salmon in 1997 began on 9 June and ended on 5 August (Figure 6). Each day fish were tagged, the fish diversion weir in the Washington-shore ladder was lowered into place in the morning to divert fish from the main portion of the ladder into the fish lab via a short section of ladder. Salmon entered the lab into a large tank with two false weirs at the top of chutes that led to a channel back to the ladder or into anesthetic tanks. As salmon swam through the water flowing over the false weirs and slid down the chutes, a person would divert the fish into the anesthetic tank by operating a hydraulic gate if the fish was one we wanted to tag, otherwise the fish entered the channel that led back into the main ladder. In this way fish were not handled prior to be anesthetized, reducing stress during the tagging process. We had no sockeye salmon mortalities during tagging, transport, or release in Tricane-methane-sulphonate (MS-222) was used to anesthetize fish at a concentration of 1 mg/l. When fish were anesthetized, they were moved to a tagging tank in a wet plastic sleeve where their length and sex (if possible) and presence of injuries, old scars, and fin clips was noted. We then outfitted fish with a transmitter that had been dipped in glycerin, by inserting it into the stomach through the mouth. The transmitter antenna was bent at the corner of the mouth and allowed to trail along the side of the fish. We used 3-volt transmitters developed and supplied by Lotek Engineering 1 that transmitted a signal every 5 s that included the frequency and code of the transmitter. The code set we used allowed us to monitor up to 17 fish on each frequency. Transmitters were powered by a lithium battery and had a rated operating life of 278 d, but usually lasted a year or more. Transmitters used in sockeye were cylindrical, 43 mm long, 14-mm in diameter and had a 47-cm long antenna, and weighed 11 g. 1 The use of this product does not constitute an endorsement by the authors. 13

19 35 3 Sockeye salmon tagged at Bonneville Dam n = 577 fish released Sockeye salmon counted at Bonneville Dam 9-Jun 16-Jun 23-Jun 3-Jun 7-Jul 14-Jul 21-Jul 28-Jul 4-Aug Figure 6. Number of sockeye salmon outfitted with radio transmitters at the Bonneville Dam adult trap (bars), and the number counted passing the dam at the counting stations (line) during the migration in We inserted a unique secondary visual implant (VI) tag into the clear tissue posterior to the eye (left usually), and a 1 mm-long piece of magnetic wire was inserted into the muscle near the dorsal fin to trigger the coded-wire detector at Lower Granite Dam. Fish were then placed in the wet sleeve and moved to the transport tank where they were held until released (usually less than 3 hours). The length of the trapping period each day depended on the number of sockeye salmon to be outfitted with transmitters and the number of fish moving up the ladder. The transport tank was a 3 gal, insulated, fiberglass tank with a large trap door on the end for fish release. Air stones in the tank bottom supplied oxygen from bottles mounted on the side of the tank. An overhead crane was used to move the transport tank in and out of the fish facility. Once trapping was finished each day, we removed diversion weir pickets from the ladder and fish in the trapping system were allowed to proceed up the ladder. The 577 sockeye salmon we tagged represented 1.2%, or 1 in 81, of the 46,665 fish counted passing Bonneville Dam during the period. We unselectively outfitted with transmitters what we believe was a near-random sample of adult sockeye salmon. The sample was not truly random because only fish passing via the Washington-shore ladder were sampled, the proportion sampled each day varied, more fish were sampled in the morning than afternoon, and no fish were sampled at night. Fish were tagged as they were trapped, and we tagged almost all fish regardless of minor injury or fin clip. We evaluated our overall sampling effort by calculating proportions of radio-tagged fish to total counts of sockeye salmon passing Bonneville Dam for consecutive 5-d blocks. Between 9 June and 5 August, the time when 99.2% of the tagged fish were 14

20 recorded passing the dam, the proportion of radio-tagged fish that passed was about 1.2% (Figure 7). Overall, however, our sampling effort through time was generally close to the overall sampling rate. In 1997, 572 (99%) of sockeye salmon outfitted with transmitters had no fin clips and 5 (1%) had adipose or ventral fin clips. Adult sockeye salmon we outfitted with transmitters in 1997 were classified as 72.3% male and 27.7% female. Fork lengths of fish tagged ranged from 38 cm to 63 cm with a median length of 49.5 cm (Figure 8). Sockeye salmon without fin clips had median fork length of 49.2 cm and those with clips had a median length of 51. cm. Sixty-seven percent of the 577 sockeye salmon tagged had no descaling, 3% less than 1%, 2% were 1-25% descaled, and <1% were more than 25% descaled. We recorded the prevalence of injuries on the heads of the fish and 96% had none, 1% had scrapes, and less than 1% had skinned areas, fungus, cuts, hook marks, or eye injuries. Sixty percent of the fish had no marks from marine mammals, 31% had fresh marine mammal scrapes, and 9% had fresh bite injuries. None of the 577 sockeye salmon had what we thought were gill net marks. Receiver and Antenna Outages During 1997, individual sequentially scanning receivers (SRX) and Yagi antennas installed at tailrace sites downstream from dams operated satisfactorily 82.1% to >99.9% of the time (mean of 92.6%, Tables 3 and 4). Tailrace receivers operated satisfactorily an average of 93.9% of the time at lower Columbia River dams and 83.8% of the time at Priest Rapids and Wanapum dams. We did not measure receiver efficiency information for sites upstream from Wanapum Dam. SRX/DSP (SRX connected to a digital scanning processor) receivers that were used to monitor the tops of ladders operated satisfactorily 83.8% to 1% of the time (mean of 95.3%). Top-ofladder receivers operated satisfactorily an average of 95.9% of the time at lower Columbia River dams and 87.7% of the time at Priest Rapids and Wanapum dams (Table 4). SRX receivers at tributary mouths operated satisfactorily 73.4% to 1% of the time (mean 95.4%). Antennas and receivers that monitored entrances to fishways and within fishways operated at similar or slightly lower rates, but data from those receivers were typically not used for the passage studies in this report. Reported receiver operations and outages include time both before and after the sockeye salmon migration in 1997, and percentages reported above do not necessarily reflect operation efficiency during the sockeye migration. Many receivers were in operation for the entire year, to monitor steelhead tagged in both 1996 and Receiver outages throughout the year occurred primarily because of power loss, receiver malfunction, vandalism, and full memory banks. In a few additional cases, receivers were operating but were not accurately recording data or were recording data incompletely. Cut or damaged antenna wires, malfunctioning receivers or downloading errors accounted for most other data gaps (Table 4). 15

21 5-day passage ratio: radio- tagged fish/fish count Total tag/count ratio: 1.2% % * * 1-Jun 15-Jun 1-Jul 15-Jul 1-Aug Figure 7. Proportion of radio-tagged sockeye salmon passing Bonneville Dam to the total counts at the dam during 5-d blocks in Blocks that include less than 2.5% of the total run noted with an asterisk. 6 5 n = 577 Median fork length = 49.5 cm Number of sockeye salmon Fork length (cm) Figure 8. Length frequency distribution of sockeye salmon outfitted with transmitters at the Bonneville adult trap in

22 Table 3. Receiver power outages and hours of operation at dams, tributaries and other fixed sites in Operation percentages do not measure data collection gaps that occurred for reasons other than power outages. * tailrace receiver; ** top-of-ladder receiver Total possible Actual Total Percent Receiver Site operation hours operation hours outage hours in operation Bonneville Dam 1BO* 8,473 8, BO* 8,76 8, BO 8,76 8, BO 8,76 7, BO 8,759 8, BO 8,76 8, BO 8,76 8, BO 8,76 8, BO 8,694 8, ABO** 4,94 4, AB , BBO 8,76 8, CBO 7,664 7, DBO 8,76 8, EBO 8,76 6, FBO 8,76 7, GBO 8,759 8, HBO 8,76 8, JBO 8,76 8, KBO 8,76 8, LBO 8,76 8, MBO 8,76 8, NBO 8,76 8, OBO 8,76 7, PBO** 7,753 7, QBO 8,76 8, RBO 8,76 6, SBO 8,76 6, TBO 6,848 6, UBO 8,76 6, VBO 6,74 6, WBO 6,776 6, XBO 6,779 6, YBO 1,381 1, ZBO 1,59 1, The Dalles Dam 1TD* 8,76 8, TD* 8,76 7,31 1, TD 1, TD** 8,76 8, ATD 6,798 6, BTD 6,798 6,798 1 CTD 6,797 6, DTD 7,733 7, ETD 6,795 6, FTD** 6,795 6, John Day Dam 1JD* 8,458 8,

23 Table 3. Continued. Total possible Actual Total Percent Receiver Site operation hours operation hours outage hours in operation JD* 8,459 8, JD** 8,76 8, JD** 8,76 8, AJD 7,439 6, BJD 6,825 6, CJD 7,329 7, DJD 8,287 7, EJD McNary Dam 1MN* 8,76 8, MN* 8,625 7,354 1, MN 8,76 7,715 1, MN 8,76 8, MN 8,76 8, MN 8,532 8, MN 8,76 8, MN 8,76 7, MN 8,759 8, AMN 8,76 8, BMN 8,626 7, CMN 8,76 8, DMN 8,76 8, EMN** 8,76 8, FMN** 8,76 8, GMN 6,822 6, HMN 6,917 6, JMN 6,583 6, KMN 6,9 6, Priest Rapids Dam 1PR* 5,617 4, PR* 5,638 4,626 1, PR 2,616 2, PR 2,647 2, PR 2,642 2, PR 2,618 2, PR 2,643 2, APR** 5,612 4, BPR 2,644 2, CPR** 5,611 4, Wanapum Dam 1WP* 5,686 4,577 1, WP* 5,61 4, WP 2,89 2, WP 2,41 2, WP 2,617 2, WP 2,617 2, WP 2,331 2, WP 2,616 2, WP 2,332 2, AWP 2,328 2, P** 5,589 5, CWP** 5,597 5,

24 Table 3. Continued. Total possible Actual Total Percent Receiver Site operation hours operation hours outage hours in operation Ice Harbor Dam 3IH 6,27 6, IH 6,272 6, IH 6,269 6, IH 3,231 3, IH 6,265 6, IH** 8,76 8, IH** 8,759 8, TIH missing outage data Lower Monumental Dam 1LM* 6,169 5, LM** 6,175 6, LM 6,147 6, LM 6,149 6, LM** 6,177 6, Little Goose Dam 1GO* 6,82 6, GO 6,31 5, GO 3,27 3, GO 2,38 2,38 1 7GO** 6,8 6, Lower Granite Dam 1GR* 8,76 7, GR 5,885 4, GR** 8,76 8, WI 2,211 2, WI 2,211 2,211 1 Tributaries Bridge of Gods (BOG) 6,79 5, Wind (WIN) 8,76 8, WNM 8,76 6,326 2, L. Wh. Salmon (LWS) 8,456 7, LWD 8,76 3,71 5, LWU 8,76 6,385 2, White Salmon (WHR) 8,76 7, WHD 6,264 3,269 2, WHU 6,356 4,29 2, Hood (HDR) 8,76 8, Klickitat (KTR) 8,76 8, Deschutes (DES) 8,76 8, DSM 8,76 6,523 2, SHF 8,76 8, John Day (JDR) 8,458 7,217 1, Umatilla (UMR) 8,76 9, Walla Walla (WWR) 8,76 8, Yakima (YAK) 8,76 8, Hanford left (HFL) 6,25 5, Hanford right (HFR) 6,26 5,7 1, Snake (SNR) 8,552 6,277 2, Clearwater (CWR) 8,123 8, SFC 5,53 5, Lochsa (LOC) 4,894 4,

25 Table 3. Continued. Total possible Actual Total Percent Receiver Site operation hours operation hours outage hours in operation Selway (SEL) 5,382 5, Imnaha (IMR) 4,72 4, Salmon (LSR) 5,528 5, SFS 3,55 3,55 1 MFS 4,666 4, USR 4,452 4, Data Collection and Processing Members of the study team downloaded data from receivers into portable computers periodically, with the frequency depending on the number of fish passing a site. Some sites were downloaded daily during the peak of the run, and some every two weeks. Each night files of downloaded data were transmitted to a computer at the NMFS lab in Seattle and added to databases. Records consisted of transmitter frequency (channel), code, date, time, power of signal received, and site. In 1997, we created databases for all the records of each fish at each dam and each species. After each day of tagging, a member of the tagging crew transmitted a file with records of fish tagged that day to the Seattle computer. When records were uploaded to the databases, the records were evaluated and good records added to the databases, and bad records were placed in a bad-record table. Bad records were those with channels and codes for fish that had not been released. As the season progressed, files of data for each dam were sent to the University of Idaho for coding by study team members. Coding of the records consisted of going through all records for a fish at a dam and assigning specific codes to identify fish activity. For example, one code would be assigned to the first record of a fish at a tailrace site downstream from a dam and another would be assigned to the last record at the tailrace site. Similarly each approach and entry into the fishways was coded, as were exits back into the tailrace and exits from the top of ladders. When all the fish had been coded for a dam, coded records were returned to Seattle and added to the databases. We had a program written to assist in coding that incorporated a decision tree that a coder would use in coding records manually. The program speeded up the coding process but still required project personnel to make final designations for behavioral codes. When all fish had been coded at each dam, all coded records for each radio-tagged salmon were combined into a file with records from tributary receivers, records of fish found by mobile trackers, and records of fish that were recaptured at weirs, hatcheries, spawning grounds, or in fisheries. Records in the file that had not been previously coded were then coded to create the general migration file, the file that contained most of the data presented in this report. Above, we referred to records of fish found by mobile trackers, and of those of fish recaptured in fisheries, at adult traps, weirs and hatcheries, and those recovered in spawning areas. Separate data files were created for mobile track records and recapture records at the University of Idaho, and data in those files were added to the databases in Seattle prior to coding the general migration file. 2

26 Table 4. Dates, duration (days) and explanation for significant gaps in data collection in 1997 by receivers and antennas. Location Start Date End Date Duration Explanation Bonneville Dam RBO and SBO 22-Jan? 24-Mar 62 Broken by ice storm ABO and CBO? 15-Feb? Unplugged JBO? 23-Apr? Memory full UBO 24-Apr 25-Apr <1 Power problem BBO and XBO 26-Apr 27-Apr <1 Memory full JBO-Antenna 1? 29-Apr? Antenna crushed NBO-Antenna 4? 29-Apr? Cable short BBO 29-Apr 2-May 4 Memory full XBO 1-May 2-May <1 Memory full EBO and FBO 7-May 15-May 8 Power outage ABO 31-May 2-Jun 2 Power disconnected ZBO 24-Jul 1-Aug 17 Replace receiver KBO? 31-Jul? Power disconnected PBO 25-Aug 2-Sep 8 Unknown ABO 22-Sep 2-Oct 1 Receiver malfunction CBO 24-Sep 29-Sep 5 Memory full DBO? 2-Oct? Memory full and power outage RBO 14-Dec 15-Dec 1 Dead battery The Dalles Dam 3TD? 6-Feb? Memory full 2TD 23-Feb 27-Feb? Receiver stolen DTD 15-Apr 19-Apr 4 Temporarily removed CTD start 6-May? Bad receiver 2TD 31-Aug 1-Oct 31 Receiver not scanning ETD 27-Aug 4-Sep 8 Antenna out of water 5TD 4-Aug 8-Oct 65 Receiver not scanning ETD 1-Aug 9-Oct 6 Line amps cut John Day Dam CJD? 1-Mar? Unplugged by contractor AJD 1-May 11-May 1 Memory full CJD? 2-Sep? Receiver not recording DJD 1-Nov 11-Nov 1 Memory full CJD 2-Nov 25-Nov 5 Unknown McNary Dam 3MN 22-Mar 1-Apr 1 Receiver running open JMN 17-Apr 22-Apr 5 Receiver running open BMN 22-Aprl 3-Apr 23 Receiver running open CMN start 14-May? Bad cable 8MN 7-May 13-May 6 Memory full 2MN 19-May 3-Jun 42 Site flooded 4MN 2-Jul? >9 No power DMN 4-Jun 12-Jun 8 Temporarily removed CMN? 4-Jun? Receiver malfunction FMN? 4-Jun? Not in DSP mode 8MN 6-Jun 11-Jun 5 Memory full 1MN 12-Jun 16-Jun 4 Loose power cord KMN 21-Jun 3-Jun 9 Unplugged EMN 14-Jul 15-Sep 6 Unknown 21

27 Table 4. Continued. Location Start Date End Date Duration Explanation JMN 2-Jul 21-Jul 1 Memory full 6MN 3-Oct 15-Oct 12 Download error 7MN 12-Oct 15-Oct 3 No power 8MN 15-Oct 23-Oct 8 Bad power strip FMN 28-Aug 15-Sep 18 Receiver problem EMN 29-Aug 15-Sep 17 Not is DSP mode KMN 15-Sep 22-Sep 7 Memory not cleared 8MN 15-Oct 23-Oct 8 No power to SRX GMN 23-Oct 31-Oct 8 Bad receiver 2MN 3-Oct 1-Nov 12 Monitor not restarted KMN 5-Nov 1-Nov 5 No power to receiver FMN 1-Nov 25-Nov 15 Loose power cord 9MN 14-Dec 22-Dec 8 Unknown Priest Rapids Dam 1PR and 2PR? 3-Apr? Dead batteries 4PR 3-Apr 1-May 1 Switched 1PR and 2PR 7-May 14-May 7 Dead batteries 6PR-Antenna 3 22-May 26-May 5 Antenna out of place 2PR 3-Jun 5-Jun 3 Not scanning 7PR 3-May 5-Jun 7 Not scanning 2PR 25-Jun 26-Jun 1 Receiver locked up 5PR 1-Jul 12-Jul 2 Memory full 8PR 11-Jul 12-Jul 1 Memory full BPR 7-Jul 16-Jul 9 Memory full CPR 6-Jul 16-Jul 1 Memory full 8PR 14-Jul 16-Jul 3 Memory full CPR 18-Jul 21-Jul 3 Memory full BPR 21-Jul 29-Jul 8 Memory full 2PR 18-Sep 23-Sep 5 Battery outage Wanapum Dam 8WP 1-May 13-May 13 Temporarily moved BWP? 4-May? Not scanning 3WP? 4-May? No power AWP-Antenna 1? 3-May? Antenna broken AWP 7-Jun 12-Jun 5 Antenna out of place 1WP 7-Jun 12-Jun 5 Receiver locked up 1WP 5-Jul 12-Jul 7 Not scanning 9WP 14-Jul 16-Jul 2 Not scanning 4WP 28-Jul 1-Aug 4 Not scanning 9WP-Antenna 5 1-Aug 6-Aug 6 Antenna broken 1WP 2-Oct 27-Oct 7 No power 1WP 23-Nov 24-Nov 1 no power Tributaries and other fixed sites LWS? 22-Apr? Memory full HDR 21-Apr 22-Apr 1 Blown breaker SNR? 24-Apr? Cut power cable LWS? 1-May? Memory full WIN 27-May 28-May 1 Low battery JDR? 2-Jun? Cows chewed cable UMR start 4-Jun? Defective receiver HFR 12-Jun 19-Jun 7 Flooded 22

28 Table 4. Continued. Location Start Date End Date Duration Explanation LSR? 6-Sep? Power out WHU 9-Aug 12-Aug 3 Dead battery WIN 19-Sep 22-Sep 3 Memory full LWD 2-Sep 22-Sep 3 Dead battery WHR 27-Sep 29-Sep 2 Memory full KTR 21-Sep 3-Sep 9 Cable cut SNR 23-Sep 9-Oct 16 Cable cut JDR 25-Aug 14-Oct 5 Receiver locked up LWD 1-Nov 3-Nov 2 Not scanning LWS 3-Nov 4-Nov 1 Antenna knocked over MFS? 13-Nov? Cable cut WHD 1-Dec 12-Dec 12 Dead battery Data for Rock Island, Rocky Reach and Wells dams provided by PUDs Statistical Methods Our sampling effort was restricted in space and time due to the location of the trapping facility and the trapping schedule (daytime only with approximately 1 d of sampling and 4 d no sampling). From early June to early August, we unselectively outfitted sockeye salmon with transmitters, but sampling rates varied (see Figure 7) due to fluctuations in the run and in tagging effort. Although not strictly random, we believe our sampling was mostly representative of the sockeye run. We did not analyze fish based on tagging schedule or release date because stocks of sockeye salmon migrated as a relatively homogenous unit to the middle Columbia River. Sockeye salmon destined for the Wenatchee and Okanogan rivers comprised 97% of fish last recorded in tributaries. Median tag dates for Wenatchee and Okanogan river stocks was 29 June. Arrival at mid-columbia River dams was also similar for both stocks. Wenatchee River stocks passed Rock Island Dam on median date of 17 July, while the median arrival date at Wells Dam for Okanogan River was 2 July. We used flow, spill, temperature, turbidity, and dissolved gas data collected at each dam to develop models on the influence of environmental conditions on sockeye passage. Most environmental conditions varied continuously at monitored dams during the study period, and several were highly autocorrelated through time and were not independent random variables (i.e. total flow and spill). We used reported daily mean values in all models, but conditions encountered by individual fish likely differed from daily means, and some fish encountered a range of conditions at a given dam. Given these statistical limitations, we believe results from modeling related to environmental conditions should be used as indicators of general trends. The study was not designed to experimentally test hypotheses related to in-river conditions (i.e. using discreet spill or flow patterns). Because sockeye salmon passage times tended to be right-skewed, we used nonparametric Wilcoxon scores and Kruskal-Wallis chi-squared (K-W X 2 ) tests (PROC NPAR1WAY, SAS Institutes Inc., 199) in time comparisons. If distributions were near normal we used parametric tests in addition to nonparametric tests. We used standard Z tests, chi-squared (X 2 ) tests of independence or X 2 goodness-of-fit tests for proportional data. All tests were two-tailed unless otherwise noted. 23

29 We initially used graphical methods for exploring univariate regression data to identify linear and non-linear trends, using loess and other data smoothing techniques. We examined residuals from univariate models for outlying data points and for nonnormality of residuals; non-normal errors were relatively common due to covariance and autocorrelation in environmental variables. Prior to building multiple regression models, we created scatterplot matrices of independent variables and identified outlying data groupings using SAS/INSIGHT. We chose forward stepwise regression to identify the most influential variables affecting passage time past projects and reservoirs, and also compared groups of models using subsets of independent variables (PROC REG, SAS Institutes Inc., 199). We chose a P cutoff value of.15 for multiple regression models because univariate correlations were relatively low in many cases. Our objectives in model building were to identify general trends and influential variables rather than to produce fully predictive models. Methods and Results For sockeye salmon tagged in 1997, we tested whether fish bound for specific tributaries or released at different sites passed Bonneville Dam via the Bradford Island and Washington-shore fishways at different than expected rates. We released 287 (49.7%) sockeye salmon with transmitters at Dodson Landing (south shore) and 29 (5.3%) at Skamania Landing (north shore). More than three-quarters of fish from both release sites were first recorded at the south-shore tailrace antenna (Table 5). About 54% of all sockeye salmon with transmitters first approached Bonneville Dam at powerhouse I, 21% first approached at powerhouse II, and 25% first approached at entrances adjacent to the spillway. Proportions were significantly different for fish from both release sites (P =.1, X 2 test). Table 5. Number of sockeye salmon with transmitters released downstream from Bonneville Dam by location, percentage 1 that were first recorded at south- and northshore tailrace receivers and percentage that passed ladders that were recorded passing the Bradford Island and Washington-shore ladders in Total ladder counts provided for comparison. Number First tailrace (%) First approach (%) Ladder passed (%) released south north PH1 PH2 spill Bradford WA-shore Sockeye salmon with transmitters All 577 (1%) Dodson 287 (5.3%) Skamania 29 (49.7%) Total sockeye salmon counts in ladders All sockeye salmon 9 June to 5 August percentage of those recorded at tailrace sites, not percentage of those released time period that radio-tagged fish were passing Bonneville Dam 24